| commit | b61a918989854abfa364b7b6849fa130c1064b09 | [log] [tgz] |
|---|---|---|
| author | Stella Laurenzo <stellaraccident@gmail.com> | Mon Mar 18 12:52:22 2024 -0700 |
| committer | GitHub <noreply@github.com> | Mon Mar 18 12:52:22 2024 -0700 |
| tree | 0f10ca550581f3ef24d7598cfac1891c25137314 | |
| parent | ee32fc7b74b4aad49fc46d67ea6c1f617416c4d4 [diff] |
Enable LTO optimization by default for runtime releases. (#16811) This is done by generalizing the primordial `IREE_SIZE_OPTIMIZED` flag into a `IREE_RUNTIME_OPTIMIZATION_PROFILE` that: * Can enable 'lto' or 'size'. * Is scoped to just the runtime targets. * Minimally does the right thing for 'size' on Linux vs just on Windows (not the goal of this patch but drops ~300KB from binary sizes when enabled). The compile time delta for a clean build of the runtime in full LTO vs regular mode was not measured precisely but is in the noise (i.e. <1m). As such, just enabling by default for Python release binaries. Others can be enabled via: `-DIREE_RUNTIME_OPTIMIZATION_PROFILE=lto`, which is recommended for benchmarking, etc. Note that this removes the use of the CMake option `IREE_SIZE_OPTIMIZED`. It was never even declared properly as an option and didn't do the same class of thing across Windows/Linux. This has been fixed and it can be enabled via `-DIREE_RUNTIME_OPTIMIZATION_PROFILE=size`. Note that as on Windows, this implies LTO. If old behavior without LTO is desired, we can add a profile for that. Progress on #898. --------- Co-authored-by: Scott Todd <scott.todd0@gmail.com>
IREE (Intermediate Representation Execution Environment, pronounced as “eerie”) is an MLIR-based end-to-end compiler and runtime that lowers Machine Learning (ML) models to a unified IR that scales up to meet the needs of the datacenter and down to satisfy the constraints and special considerations of mobile and edge deployments.
See our website for project details, user guides, and instructions on building from source.
IREE is still in its early phase. We have settled down on the overarching infrastructure and are actively improving various software components as well as project logistics. It is still quite far from ready for everyday use and is made available without any support at the moment. With that said, we welcome any kind of feedback on any communication channels!
See our website for more information.
IREE is licensed under the terms of the Apache 2.0 License with LLVM Exceptions. See LICENSE for more information.